functions of mammalian TOR complex 2
Inauguraldissertation
zur
Erlangung der Würde eines Doktors der Philosophie vorgelegt der
Philosophisch-Naturwissenschaftlichen Fakultät der Universität Basel
von
Nadine Cybulski
aus Andermatt (UR)
Basel, 2010
Genehmigt von der Philosophisch-Naturwissenschaftlichen Fakultät auf Antrag von Prof. Dr. Michael N. Hall und Prof. Dr. Markus A. Rüegg.
Basel, den 28. April 2009
Prof. Dr. Eberhard Parlow Dekan
Table of Contents
SUMMARY 2
INTRODUCTION 5
THE TOR SIGNALING PATHWAY 6
THE TOR KINASE 7
TOR COMPLEX 1 8
TOR COMPLEX 2 13
MTOR ASSOCIATED DISEASES 23
ADIPOSE TISSUE 25
FAT STORAGE AND RELEASE 26
ENDOCRINE FUNCTIONS 27
ADIPOSE TISSUE ASSOCIATED DISEASES 29
CELL MIGRATION 30
AIMS OF THE THESIS 31
RESULTS 33
PART 1: PHENOTYPIC CHARACTERIZATION OF RICTOR ADIPOSESPECIFIC KNOCKOUT MICE 35
ABSTRACT 37
INTRODUCTION 37
RESULTS 39
DISCUSSION 43
METHODS 44
FIGURES AND TABLES 47
ADDITIONAL RESULTS 54
PART 2: MTORC2 IS REQUIRED FOR CELL MIGRATION 61
ABSTRACT 63
INTRODUCTION 63
RESULTS 65
DISCUSSION 70
MATERIALS AND METHODS 72
FIGURES 75
FINAL CONCLUSIONS & OUTLOOK 81
REFERENCES 86
APPENDIX: ADIPOSE MTORC1 CONTROLS ENERGY HOMEOSTASIS 103
ACKNOWLEDGEMENTS 116
Summary
Target of rapamycin (TOR) is the main controller of cell growth and metabolism in response to nutrients, growth factors and the cellular energy status. TOR is a serine/threonine kinase conserved from yeast to mammals and is found in two functionally and structurally distinct multi-protein complexes named TOR complex 1 (TORC1) and TORC2. Mammalian TORC1 (mTORC1) contains mTOR, mLST8, raptor and PRAS40, while mTORC2 contains mTOR, mLST8, rictor, mSin1, and PRR5. mTORC2 is activated in response to growth factors, such as insulin and insulin-like growth factor 1 (IGF1), and its main functions involve the regulation of actin cytoskeleton dynamics and phosphorylation of several AGC kinases in their hydrophobic motif. TORC1 is directly inhibited by the immunosuppressant and anti-cancer drug rapamycin, whereas TORC2 is not. Thus, use of rapamycin provides a simple and straightforward method to specifically study the TORC1 signaling branch. There is no known TORC2-specific inhibitor, so genetic manipulation is required to study its biological function(s).
This thesis describes new in vivo and in vitro functions of mTORC2. The first part deals with the in vivo function of mTORC2 in adipose tissue. The adipose tissue, in addition to its function as a long-term fat storage depot, also has endocrine functions, plays an important role in the regulation of whole body glucose and lipid metabolism and is one of the most insulin-responsive tissues in the body. To study mTORC2 function in adipose tissue, we have generated mice that lack the mTORC2-essential component rictor specifically in adipose tissue. Phenotypic characterization revealed the unexpected finding that these mice were larger due to an increase in lean tissue mass and that they had elevated serum IGF1 levels. Furthermore, the knockout mice were hyperinsulinemic, but glucose tolerant.
Overall, these findings suggest an important role for adipose mTORC2 in controlling full body growth and whole body glucose metabolism.
The second part of this thesis describes a new in vitro function of mTORC2 in fibroblasts. We have taken advantage of the raptor and rictor floxed mice to isolate mouse embryonic fibroblasts (MEFs), which were then used to establish inducible raptor and rictor knockout MEF cell lines. After initial characterization of these two cell lines, a deeper analysis of the role of mTORC2 in the actin-mediated process of cell migration was performed. We have found that mTORC2 is required for cell migration and for regulating the activity of the Rho GTPases Rac1, Cdc42, and RhoA. We have extended this study by
showing that mTORC2-dependent cell migration is also required in oncogenic cells, which suggests that mTORC2 could have an important function in the development of cancer and metastasis.
Introduction
The TOR signaling pathway
Rapamycin is an antifungal metabolite that was initially isolated from soil samples from the Easter Island (locally known as Rapa Nui) in the 1970s and that was found to have inhibitory effects on cell proliferation of mammalian cells. Further studies led to the identification of the TOR (target of rapamycin) kinase in the budding yeast Saccharomyces cerevisiae (Heitman et al., 1991). Yeast cells bearing mutations in the TOR genes were resistant to rapamycin, which normally forms a complex with the peptidyl-prolyl cis/trans isomerase FKBP12 to bind to and inhibit the TOR kinase. TOR is a Ser/Thr kinase and is conserved in every eukaryote examined so far, including yeasts, algae, slime mold, plants, worms, flies and mammals (Wullschleger et al., 2006). Recently, TOR was also identified in the protozoan parasite Trypanosoma brucei (Barquilla et al., 2008). The main function of TOR is the regulation of cell growth, and its function is critical for a cell or organism since disruption of the gene is lethal in all eukaryotes (Kunz et al., 1993; Long et al., 2002; Menand et al., 2002; Oldham et al., 2000). For example, mice deficient for mammalian TOR (mTOR) die very early during embryonic development (Gangloff et al., 2004; Murakami et al., 2004). The importance of a functional TOR signaling pathway is further underscored by the high incidence of the TOR pathway being involved in human diseases. Dysfunctional TOR signaling is associated with many forms of cancer, and is linked to diseases such as tuberous sclerosis complex (TSC) or diabetes.
TOR is a central controller of cell growth and is regulated in response to nutrients, growth factors and cellular energy status. To fulfill its function within the cell TOR is found in two structurally and functionally distinct multiprotein complexes, TOR complex 1 (TORC1) and TORC2. In mammals, rapamycin-sensitive TORC1 (mTORC1) contains mTOR, mLST8, and its specific components raptor and PRAS40 (Fonseca et al., 2007; Hara et al., 2002; Kim et al., 2002; Kim et al., 2003a; Loewith et al., 2002; Oshiro et al., 2007; Sancak et al., 2007;
Thedieck et al., 2007; Vander Haar et al., 2007; Wang et al., 2007). mTORC2 is rapamycin insensitive and contains mTOR, mLST8 and the specific components rictor, mSin1, and PRR5 (Frias et al., 2006; Jacinto et al., 2006; Jacinto et al., 2004; Pearce et al., 2007;
Sarbassov et al., 2004; Thedieck et al., 2007; Woo et al., 2007; Yang et al., 2006).
In the following, TOR itself and the individual complexes, along with their upstream regulators and downstream substrates, will be described in more detail.
The TOR kinase
The TOR genes, TOR1 and TOR2, were initially identified in the budding yeast S. cerevisiae (Heitman et al., 1991), where they share 67% primary sequence identity. While some yeasts and the protozoan parasite T. brucei have two TOR genes, higher eukaryotes only have a single TOR gene, but they all encode for a very large protein of ∼ 280 kDa in size. The TOR proteins share high sequence homology among all eukaryotes and they belong to the same kinase family known as phosphatidylinositol kinase-related kinase (PIKK) family (Abraham, 2001; Keith and Schreiber, 1995). Although the TOR kinase domain resembles the catalytic domain of the lipid kinases phosphatidylinositol 3-kinases, TOR is a serine/threonine protein kinase and is the founding member of the PIKK family that includes the ATM, ATR, TOR, SMG-1, TRRAP and DNA-PK subfamilies of PIKKs (Abraham, 2004).
The TOR proteins are composed of numerous highly conserved domains (Figure I-1).
The amino-terminal part contains a large stretch of tandem HEAT repeats. This motif is common among Huntingtin, Elongation factor 3, the A subunit of PP2A and TOR. The HEAT repeats are composed of 40-50 amino acids and form a structure of two tandem anti-parallel α-helices that are thought to mediate protein-protein interactions (Perry and Kleckner, 2003).
Towards the carboxy-terminal domain the HEAT repeats are followed by a moderately conserved FAT (FRAP, ATM and TRRAP) domain, which is a shared domain of the PIKK family and is always found together with the highly conserved FATC (FAT C-terminus) domain at the very carboxy-terminal end of the protein. The FAT and FATC domains are believed to interact with each other to modulate kinase activity. Next to the FAT domain is the FRB (FKBP-Rapamycin Binding) domain that allows binding of the FKBP-rapamycin complex, thereby leading to allosteric inhibition of the kinase. The kinase domain of the protein lies next to the FRB domain. Furthermore, a repressor domain was identified in mTOR and comprises a region of 20 amino acids (2430-2450) located directly upstream of the FATC domain. Deletion of this region results in increased TOR kinase activity (Sekulic et al., 2000). This repressor domain also contains two phosphorylation sites: Thr2446 is phosphorylated by AMP-dependent kinase (AMPK) (Cheng et al., 2004) and Ser2448 is phosphorylated by the TOR substrate S6K1 (Chiang and Abraham, 2005; Holz and Blenis, 2005). However, the physiological meaning of these phosphorylation sites is not well understood. Additionally, an autophosphorylation site was identified at Ser2481, but the physiological relevance is also not well defined (Peterson et al., 2000). Only recently, it was suggested that this phosphorylation is specific to only one of the two complexes and resembles the kinase activity of the complex (Copp et al., 2009). A schematic summary of the various domains within the TOR kinase is depicted in Figure I-1.
The abundance of several putative protein-protein interaction domains clearly suggested that TOR might associate with many cellular proteins. This assumption was also supported by gel filtration experiments demonstrating that TOR consistently migrated in a 2 MDa complex (Kim et al., 2002; Loewith et al., 2002). These ideas led to the identification of several binding partners of TOR that were subsequently found to define the two functionally distinct multiprotein complexes, TOR complex 1 (TORC1) and TORC2. The complexes will be described in more detail in the next part.
Figure I-1. Illustration of the domain structure of mammalian TOR.
For more details see text.
TOR complex 1
Composition and localization
In yeast, TORC1 is comprised of either TOR1 or TOR2, LST8 (lethal with sec thirteen), KOG1 (kontroller of growth), and TCO89 (89-kDa subunit of TOR complex one) (Loewith et al., 2002; Reinke et al., 2004). TCO89 is the only non-essential component of yeast TORC1 and no homologue was found in higher eukaryotes. Deletion of any other member of the complex is lethal in yeast (Heitman et al., 1991; Loewith et al., 2002). Interestingly, only when assembled with the TORC1 components, the FKBP12-rapamycin inhibitory complex can bind to the FRB domain and block the activity of TOR.
Mammalian TORC1 (mTORC1) consists of mTOR, mLST8 and the KOG1 homologue raptor (Hara et al., 2002; Kim et al., 2002; Kim et al., 2003a; Loewith et al., 2002). A full body knockout of any component of mTORC1 in mice results in early embryonic lethality (Gangloff et al., 2004; Guertin et al., 2006). mLST8 (previously also known as GβL) is a 36 kDa protein that consists entirely of seven WD40 repeats (about 40 amino acids with
conserved W and D forming four anti-parallel β strands) allowing protein-protein interaction (Kim et al., 2003a). The molecular function of mLST8 within mTORC1 is still ambiguous.
Initial studies suggested that the constitutive binding of mLST8 to mTOR was important for the activity of mTORC1 (Kim et al., 2003a). However, more recent findings in mLST8-/- mouse embryonic fibroblasts (MEFs) revealed that mLST8 is dispensable for mTORC1 function (Guertin et al., 2006). Also, in S. cerevisiae no clear data could be obtained on the role of LST8 in regulating TORC1. raptor is a 150 kDa protein and contains a highly conserved amino-terminal region followed by three HEAT repeats and seven WD40 repeats.
raptor is a positive regulator of mTOR activity and functions as a scaffold protein to couple mTOR to its substrates (Hara et al., 2002; Kim et al., 2003a; Schalm et al., 2003). Inhibition of mTORC1 by rapamycin results in the disruption of raptor binding to mTOR (Kim et al., 2002; Oshiro et al., 2004).
Despite several approaches in various labs, the localization of TORC1/mTORC1 within the cell is not clearly defined. However, a common finding of all studies in yeast suggests that TORC1 associates with membranes. Binding of TORC1 to the plasma membrane, vacuolar membrane and also endosomal membranes was demonstrated (Araki et al., 2005; Cardenas and Heitman, 1995; Chen and Kaiser, 2003; Kunz et al., 2000; Reinke et al., 2004; Sturgill et al., 2008; Wedaman et al., 2003), but the physiological significance of TORC1 association to distinct membranes is not understood, as well as the question whether TORC1 localization is a regulated and dynamic process. In mammalian cells the localization of mTOR to endomembranes is consistent with the findings in yeast. But mTOR localization also remains ambiguous as individual studies proposed different compartmental localization of mTOR, including mitochondrial, endoplasmic reticulum and Golgi apparatus membranes (Desai et al., 2002; Drenan et al., 2004). One study even suggests a nuclear localization of mTOR (Bachmann et al., 2006). Recently, new findings propose a regulated localization of mTOR to specific endomembranes upon activation by amino acids (Sancak et al., 2008).
Upstream regulators of mTORC1
mTORC1 integrates three major inputs to regulate catabolic and anabolic processes that collectively determine cell growth and metabolism. Growth factors (insulin/insulin-like growth factor) and nutrients (amino acids) activate mTORC1, while a low cellular energy status inhibits mTORC1 (also see Figure I-2).
Activation by insulin occurs via the well-established insulin/PI3K/Akt pathway. Insulin binding to its receptor activates a cascade of phosphorylation and recruitment events leading to the phosphorylation and activation of Akt (also known as PKB), which then can activate
mTORC1 via two pathways. The main pathway involves an inhibitory phosphorylation of the TSC1/TSC2 (tuberous sclerosis complex 1 and 2) complex, a GTPase activating protein (GAP) towards the Rheb GTPase (Garami et al., 2003; Inoki et al., 2003a; Inoki et al., 2002;
Manning et al., 2002; Potter et al., 2002; Tee et al., 2003; Zhang et al., 2003). When the TSC1/TSC2 complex is inactive Rheb is in its active GTP-bound form and can directly bind and activate mTORC1 (Long et al., 2005a; Long et al., 2005b; Smith et al., 2005). However, the molecular mechanism of how Rheb activates mTORC1 is not clear. Akt can further activate mTORC1 by an inhibitory phosphorylation of the protein PRAS40 (Proline-rich Akt substrate) that otherwise binds raptor and thereby inhibits mTORC1 (Fonseca et al., 2007;
Oshiro et al., 2007; Sancak et al., 2007; Thedieck et al., 2007; Vander Haar et al., 2007;
Wang et al., 2007).
Figure I-2: Overview of the mTOR signaling pathway. For more details see text.
Regulation of mTORC1 by the cellular energy status also involves the TSC1/TSC2 complex.
If energy levels (glucose availability) are low within the cell, the AMP/ATP ratio is increased and sensed by AMPK. AMP directly binds to and allosterically activates AMPK to allow full activation of AMPK by the tumor suppressor LKB1 (Shaw et al., 2004b). LKB1
phosphorylates and activates AMPK, which in turn phosphorylates TSC2 and activates the TSC1/TSC2 complex causing inhibition of mTORC1 (Corradetti et al., 2004; Inoki et al., 2003b; Shaw et al., 2004a). Overall, active AMPK turns on ATP-generating catabolic pathways, such as fatty acid oxidation and glycolysis, and shuts off ATP-consuming anabolic processes, such as translation and fatty acid synthesis.
The major nutritional input that regulates mTORC1 activity is the supply of amino acids.
Amino acids, in particular leucine and arginine, are essential for mTORC1 activation, since insulin alone is not sufficient to activate mTORC1 (Hara et al., 1998). The molecular mechanism how amino acids activate mTORC1 is not fully understood, however, several studies investigated single steps of mTORC1 activation to narrow down the proteins involved in amino acid activation of mTORC1. In cells lacking either TSC1 or TSC2 the mTORC1 pathway remains sensitive to amino acid starvation, suggesting that amino acids activate mTORC1 downstream of the TSC1/TSC2 complex (Smith et al., 2005). Further studies suggested that amino acids regulate Rheb binding to mTORC1 (Long et al., 2005b). Only recently, a new regulator of the mTORC1 signaling branch was identified. The Rag proteins belong to the family of small GTPases and were shown to regulate mTORC1 activity in the presence of amino acids (Kim et al., 2008; Sancak et al., 2008). Amino acids increase the binding of active Rag to mTORC1 thereby promoting the intracellular localization of mTOR to the proximity of its activator Rheb. Overexpression of a constitutively active mutant of Rag rendered the mTORC1 branch insensitive to amino acids.
Several other studies proposed another mechanism how amino acids signal to mTORC1 independent of the TSC1/TSC2-Rheb-Rag branch and they include the positive mTORC1 regulators hVps34 (vacuolar protein sorting 34) (Byfield et al., 2005; Nobukuni et al., 2005), a class III PI3 kinase, and MAP4K3 (mitogen-activated protein kinase kinase kinase kinase 3) (Findlay et al., 2007).
Besides the three major inputs – nutrients, growth factors and cellular energy status – that regulate mTORC1 activity, additional pathways can influence mTORC1 signaling. These include positive regulation via the Wnt signaling pathway (Inoki et al., 2006), mitogenic activation by the lipid second messenger phosphatidic acid (PA) (Fang et al., 2003; Fang et al., 2001; Ha et al., 2006; Kam and Exton, 2004; Sun et al., 2008) and negative regulation via cellular stresses such as hypoxia through REDD1 and 2 (Regulated in Development and DNA Damage Response genes 1 and 2) (Brugarolas et al., 2004; DeYoung et al., 2008;
Reiling and Hafen, 2004).
Downstream effectors of mTORC1
In mammals, mTORC1 regulates various cellular processes including ribosome biogenesis, translation, transcription, autophagy, and mitochondrial metabolism (Soulard and Hall, 2007).
Despite influencing a broad range of cellular processes only two direct substrates for mTORC1 have been identified so far: S6K (p70 S6 kinase) and 4E-BP1 (eIF4E-binding protein 1, also known as PHAS-I) (Brunn et al., 1997; Burnett et al., 1998). Both proteins contain a TOS (TOR signaling) motif, a five amino acid sequence that allows binding of the substrates to raptor, and which is crucial for mTOR-dependent phosphorylation (Nojima et al., 2003; Schalm and Blenis, 2002; Schalm et al., 2003).
S6K belongs to the family of AGC kinases (protein kinases A, G and C) and once activated phosphorylates the ribosomal protein S6 to activate translation. S6K contains several mTOR phosphorylation sites, but the main site required for S6K activation is Thr389, which is located in the hydrophobic motif (Pearson et al., 1995). Phosphorylation in the hydrophobic motif is required for interaction of S6K with phosphoinostide-dependent kinase 1 (PDK1) and subsequent phosphorylation and full activation by PDK1 at Thr229 in the activation loop (Alessi et al., 1998; Pullen et al., 1998). The hydrophobic motif and the activation loop, along with the mode of activation, are common to all AGC kinases. Several other rapamycin- sensitive phosphorylation sites in S6K were also described (Isotani et al., 1999; Pearson et al., 1995; Saitoh et al., 2002).
In addition to being a downstream effector of mTORC1 signaling S6K also has an important regulatory function on the upstream insulin signaling pathway, which is defined as the negative feedback loop. Upon activation of mTORC1, activated S6K phosphorylates IRS1 at multiple inhibitory sites thereby promoting degradation of IRS (Harrington et al., 2004; Haruta et al., 2000; Shah et al., 2004; Tremblay et al., 2007; Ueno et al., 2005; Um et al., 2004). As a consequence, further Akt signaling by insulin is attenuated and cells are in an insulin-resistant state.
4E-BP1 is a small protein (12 kDa) and negatively regulates translation initiation.
Hypophosphorylated 4E-BP1 binds to and sequesters eIF4E (eukaryotic initiation factor 4E).
mTORC1 phosphorylates 4E-BP1 at multiple sites and thereby causes the dissociation from eIF4E (Gingras et al., 1999; Gingras et al., 2001; Mothe-Satney et al., 2000). Upon release from 4E-BP1, eIF4E recruits additional factors, which will finally result in the assembly of the small ribosomal subunit and the recruitment to the mRNA to initiate translation.
Overall, upon activation mTORC1 positively regulates translation initiation via activating phosphorylation of S6K and inhibitory phosphorylation of 4E-BP1.
TOR complex 2
The following part on TOR complex 2 is part of a manuscript prepared for an invited review for publication in TiBS (Trends in Biological Sciences). It covers the major findings on TOR complex 2 in budding yeast, fission yeast, slime mold, worm, flies and mammals.
TORC2 in budding yeast
TOR was initially identified in the budding yeast Saccharomyces cerevisiae, in a genetic selection for spontaneous rapamycin resistant mutants (Heitman et al., 1991). In contrast to other eukaryotes, yeast (budding and fission yeast) contains two TOR genes, TOR1 and TOR2. The existence of two TORs in yeast facilitated the study of TOR signaling as it initially helped to identify two separate TOR signaling branches. Biochemical studies later demonstrated the existence of two functionally distinct TOR complexes that correspond to and thereby confirm the two previously identified branches. Whereas rapamycin-sensitive TORC1 contains either TOR1 or TOR2, rapamycin-insensitive TORC2 contains only TOR2 of the two TORs. TORC2 consists also of AVO1, AVO2, AVO3, LST8 and BIT61 (Loewith et al., 2002; Reinke et al., 2004) (Figure I-3a). AVO1, AVO3 and LST8 are essential, conserved proteins required for kinase activity. In contrast, AVO2 and BIT61 are not essential, and no clear homologous counterparts have been identified so far in higher eukaryotes, although BIT61 and the mammalian TORC2 (mTORC2) component proline-rich protein 5 (PRR5) share low sequence similarity (Hayashi et al., 2007; Woo et al., 2007). Studies investigating the molecular organization of TORC2 in yeast revealed that TORC2 is oligomeric, likely a TORC2-TORC2 dimer (Wullschleger et al., 2005).
Even before the two TOR complexes were identified, TOR2, but not TOR1, was known to regulate the cell cycle-dependent polarization of the actin cytoskeleton (Schmidt et al., 1996), thereby implicating TOR2, and hence later on TORC2, in the spatial control of yeast cell growth. Further studies showed that the aberrant depolarization of the actin cytoskeleton in TORC2 temperature sensitive mutants could be suppressed by hyperactivation of the cell wall integrity pathway, which involves Protein Kinase C1 (PKC1).
In fact, several genetic studies revealed that PKC1, via the Rho-like GTPases RHO1 and RHO2 and their GDP/GTP exchange factor ROM2, mediates TORC2 signaling to actin organization (Bickle et al., 1998; Helliwell et al., 1998a; Helliwell et al., 1998b; Loewith et al., 2002; Schmidt et al., 1997). However, further molecular details on the activation of this pathway required the identification of TORC2 substrates. In 2005, Kamada et al. identified the first yeast TORC2 substrate, Yeast Protein Kinase 2 (YPK2) (Kamada et al., 2005).
YPK2 is an AGC kinase and closely related to mammalian serum and glucocorticoid induced protein kinase-1 (SGK1). TORC2 activates YPK2 by directly phosphorylating Ser641 and Thr659 in the turn and hydophobic motifs, respectively (Figure I-3a). Recent results showed that TORC2 activity is also required for the phosphorylation of the turn motif in PKC1 (Facchinetti et al., 2008), although it is unclear whether TORC2 phosphorylates PKC1 directly. In addition to YPK2, Slm proteins, which bind phosphatidylinositol-4,5-bisphosphate (PIP2), have been characterized as direct TORC2 substrates. Slm1 and Slm2 can regulate actin organization independently of YPK2 (Fadri et al., 2005). However, a constitutively active mutant of YPK2 suppresses the lethality provoked by the complete loss of TORC2 (Kamada et al., 2005), suggesting that YPK2 is the main TORC2 effector. Most TORC2 mediated functions, including actin remodeling, are now believed to be mediated via YPK2.
Aronova et al. (Aronova et al., 2008) recently described a new function for TORC2.
They showed that TORC2, via YPK2, controls the sphingolipid biosynthetic pathway and hence mediates de novo ceramide biosynthesis. Other studies have shown that YPK2 and its homologue YPK1 are also involved in regulating eisosome assembly and turnover in a sphingolipid-dependent manner (Luo et al., 2008). Eisosomes are protein complexes near the plasma membrane involved in the early steps of endocytosis. Whether TORC2 is also directly involved in the regulation of endocytosis remains to be confirmed. Interestingly, GFP-tagged TOR2 localizes to punctate structures in the proximity of the plasma membrane, which resemble eisosomes (Sturgill et al., 2008).
Thus, the downstream effectors and functions of TORC2 in S. cerevisiae are coming into focus, the upstream regulators of TORC2 remain completely unknown.
TORC2 in fission yeast
Schizosaccharomyces pombe, like S. cerevisiae, has two TOR homologues, Tor1 and Tor2.
However, and this easily leads to confusion, fission and budding yeast TOR proteins are numbered in the opposite way, as the S. pombe proteins were named based on order of discovery rather than based on function. Budding yeast TOR2 is the sole TOR protein found in TORC2, whereas in fission yeast Tor1 is the main determinant of TORC2 (Matsuo et al., 2007), suggesting that S. pombe Tor1 is functionally equivalent to S. cerevisiae TOR2. S.
pombe TORC2 is composed of Tor1, Sin1, Ste20, Wat1 (also known as Pop3) and Bit61 (Figure I-3b). S. pombe Tor1 is not essential for normal growth but is required for survival under stress conditions, proper G1 arrest, and sexual development (Kawai et al., 2001;
Weisman and Choder, 2001). Surprisingly, in contrast to other organisms, rapamycin has no effect on normal growth in S. pombe. Initially, rapamycin was found to affect sexual development; later experiments showed that Tor1-mediated amino acid uptake in fission
yeast is also rapamycin sensitive (Weisman et al., 1997; Weisman et al., 2001; Weisman et al., 2005). As rapamycin blocks Tor1-mediated functions, it was believed that rapamycin in fission yeast inhibits TORC2 function. However, Petersen and Nurse recently showed that rapamycin can also inhibit Tor2, and hence TORC1, but in a nutrient-dependent manner (Petersen and Nurse, 2007). Currently, it remains unclear why rapamycin does not arrest fission yeast growth as in other eukaryotes. Furthermore, the molecular details on how rapamycin inhibits TORC2 and possibly TORC1 are not well understood.
To identify potential substrates of fission yeast Tor1, Matsuo et al. performed a high- copy suppressor screen of a Tor1 sterility mutant and isolated Gad8 as a potential candidate (Matsuo et al., 2003). Gad8 is a Ser/Thr kinase belonging to the AGC kinase family and is the fission yeast homologue of YPK2 in budding yeast and SGK1 in other organisms.
Matsuo et al. demonstrated that the activity and phosphorylation status of Gad8 depend on Tor1 activity. On a molecular level, they showed that the critical Tor1 phosphorylation sites in Gad8 are Ser527 and Ser546, the turn and hydrophobic motif sites, respectively (Figure I- 3b). These findings contributed very early to the idea that TOR complexes regulate many AGC kinases.
TORC2 in Dictyostelium
Dictyostelium discoideum is a powerful model organism to study mechanisms of cell movement and chemotaxis. This slime mold is a unicellular eukaryotic organism that, upon starvation, forms multicellular aggregates. This developmental program depends on chemotaxis toward a high extracellular level of cyclic AMP secreted by neighboring cells.
Lee et al. showed that the underlying mechanism of cell movement and aggregate formation depends on TORC2 (Lee et al., 2005). TORC2 in D. discoideum comprises Tor, Lst8, Rip3 (AVO1 in S. cerevisiae) and Pia (AVO3 in S. cerevisiae) (Figure I-3c), and cells deficient for any of these components lose speed, cell polarity and directionality, i.e., they display an overall defect in chemotaxis. Furthermore, Lee et al. demonstrated that cells lacking any TORC2 component exhibit reduced PKBA and PKBR1 activity, two Akt (also known as protein kinase B) homologues and AGC kinase family members that are required to fully activate the chemotactic response.
Chemoattractant signaling triggers several cellular responses. One well characterized response is the production of phosphatidylinositol 3,4,5-triphosphate (PIP3) at the leading edge of motile cells. It is well established that PIP3 production is an important step in regulating chemotaxis; however, chemotaxis still occurs in the absence of PIP3 (Chen et al., 2003; Hoeller and Kay, 2007), and a recent study provided insight on the molecular pathway regulating PIP3-independent chemotaxis. Kamimura et al. (Kamimura et al., 2008)
showed that TORC2 is activated in a PIP3-independent manner by a heterotrimeric G protein and by cytosolic Ras GTPases. This signaling leads to activation of the PKBs, mainly PKBR1, which in turn phosphorylate several downstream targets to ultimately regulate directed cell movement. These findings suggest a possibly direct regulation of TORC2 by Ras. This idea is further supported by the presence of a Ras-binding domain in Rip3 (Ras- interacting protein-3), the mSin1 homologue in D. discoideum, which, when mutated, causes an impaired chemotactic response (Lee et al., 2005). Kamimura et al. also confirmed that TORC2 activity is required to phosphorylate the hydrophobic motif in PKBA and PKBR1 and thus to activate these AGC kinases (Lee et al., 2005). Overall, these studies in D.
discoideum underscore the importance of TORC2 in actin remodeling and cell movement, and could provide insight on the role of TORC2 in actin organization in other organisms. D.
discoideum might also be a particularly valuable system for unraveling the upstream regulation of TORC2. So far, only very little is known about the upstream regulation of TORC2 in other organisms, and it would be of interest to know whether Ras-mediated activation of TORC2 is conserved in other organisms.
TORC2 in worms and flies
Recently, two independent studies have demonstrated that TORC2 is also present in the nematode Caenorhabditis elegans (Jones et al., 2009; Soukas et al., 2009). Both studies identified CeRictor in a screen for mutants with altered lipid storage and showed that loss-of- function mutants in CeRictor are viable, but developmentally delayed with a reduced overall body size (Figure I-3d). Increased fat storage in the mutant worms suggests an important role for TORC2 in regulating fat metabolism. Interestingly, an sgk1 null mutant, but not a mutant with impaired AKT signaling, phenocopies a CeRictor mutant, and a constitutively active SGK1 suppresses a CeRictor mutation. Thus, TORC2 in worms appears to signal mainly through SGK1. This, combined with findings in yeast suggesting that the SGK1 homolog YPK2 is the main TORC2 effector, lends doubt to whether Akt is the primary effector of TORC2 in other eukaryotes. Loss of mTORC2 in mammals has only little effect on Akt activity (see below).
Similar to TORC2 mutants in worms, rictor and sin1 mutants of the fly Drosophila melanogaster are viable, but reduced in body size (Hietakangas and Cohen, 2007; Lee and Chung, 2007). Also in flies, TORC2 is the main kinase phosphorylating Akt at the hydrophobic motif (Figure I-3e). Although loss of TORC2 normally causes only a mild growth
Figure I-3: The conserved TORC2 pathway. TORC2 structure and function are conserved in (a) budding yeast (S. cerevisiae), (b) fission yeast (S. pombe), (c) slime mold (D.
discoideum), (d) worms (C. elegans), (e) flies (D. melanogaster) and (f) mammals. In all organisms, TORC2 is composed of the TOR kinase (orange), LST8 (yellow), AVO3 or its homologue (green), and AVO1 or its homologue (brown). In S. cerevisiae, S. pombe and mammals, additional TORC2 components were identified (white). All shown TORC2 substrates (blue) are members of the AGC kinase family. The AGC kinases share a conserved mode of regulation involving phosphorylation of their hydrophobic motif (red), turn motif (pink), and activation loop (gray). TORC2 phosphorylates the hydrophobic motif in all shown kinases. TORC2 has so far been shown to phosphorylate the turn motif, directly or indirectly, in YPK2, Gad8, and mammalian Akt and PKCα. PDK1, which is also conserved in all organisms shown, phosphorylates the activation loop.
defect in flies, Hietakangas and Cohen (Hietakangas and Cohen, 2007; Lee and Chung, 2007) made the interesting observation that loss of TORC2 strongly inhibits hyperplasia caused by elevated phosphatidylinositol-3-kinase (PI3K) signaling, suggesting that TORC2- regulated Akt activity might be more important in conditions of elevated Akt signaling.
TORC2 in mammals
Mammalian TORC2 was identified in 2004. At that time, TORC2 was known to consist of mTOR, mLST8 and the TORC2-specific component rictor (Jacinto et al., 2004; Sarbassov et al., 2004). More recently, two additional complex-specific components were identified - the AVO1 homologue mSin1 (Frias et al., 2006; Jacinto et al., 2006; Yang et al., 2006), and the BIT61 family members PRR5 and PRR5L (also known as Protor1 and Protor2) (Pearce et al., 2007; Thedieck et al., 2007; Woo et al., 2007). Except for PRR5 and PRR5L, all mTORC2 components are essential and knockout of any one of them in mice results in developmentally delayed embryos that die in midgestation around embryonic day E10.5 (Guertin et al., 2006; Jacinto et al., 2006; Shiota et al., 2006; Yang et al., 2006).
Similar to TORC2 in S. cerevisiae, mTORC2 cannot be directly inhibited by rapamycin. However, in a few cell lines, prolonged rapamycin treatment can inhibit mTORC2 activity indirectly (Sarbassov et al., 2006). In their study, Sarbassov et al. showed that long- term rapamycin treatment prevents de novo mTORC2 assembly and thereby inhibits mTORC2 activity, but only in a few of the many cell lines examined. The effect of rapamycin on mTORC2 assembly appears to be due to rapamycin binding free mTOR and to an indirect consequence of rapamycin’s inhibition of protein synthesis and thus the synthesis of new mTOR.
A recent study by Copp et al. (Copp et al., 2009) showed that mTOR is phosphorylated in an mTORC-specific manner. mTOR phosphorylation at Ser2448 is predominantly, but not exclusively, associated with mTORC1 whereas mTOR in mTORC2 is specifically phosphorylated at Ser2481. Copp et al. suggest that the mTORC2 specific phosphorylation at Ser2481 could be used as a biomarker for mTORC2 sensitivity to rapamycin. However, to date, the functional importance of these phosphorylation sites is completely unknown, and it is important to point out that these sites should always be used along with other complex-specific readouts to definitively specify complex activation.
The first function ascribed to mTORC2, based on the previously known function of TORC2 in yeast, is the regulation of the actin cytoskeleton. Knockdown of mTORC2-specific components in cultured cells results in alteration of the actin cytoskeleton. Furthermore, also as in yeast, it was suggested that mTORC2 signals to the actin cytoskeleton via
RhoGTPases and PKC (Jacinto et al., 2004; Sarbassov et al., 2004). However, while two research groups have independently observed an altered actin cytoskeleton upon knockdown of mTORC2-specific components, opposite phenotypes were observed. Jacinto et al. observed that mTORC2 is required for cell spreading and actin fiber assembly. In contrast, the findings by Sarbassov et al. suggest that loss of mTORC2 promotes actin fiber assembly. More recently, a role for mTORC2 in regulating the actin cytoskeleton was questioned when no obvious alterations in the actin cytoskeleton were observed in embryonic fibroblasts derived from rictor knockout mice (Guertin et al., 2006; Shiota et al., 2006). The apparent discrepancies on mTORC2-mediated actin regulation could possibly be related to the different systems studied. The early knockdown studies looked at actin changes immediately after loss of mTORC2, but used different cell lines, whereas the subsequent studies looked at knockout cells permanently deficient for mTORC2. One possible explanation could be that cells constitutively lacking mTORC2 might adapt by using other mechanisms to regulate actin cytoskeletal organization. Several other studies using different approaches have supported a role for mTORC2 in actin-regulated processes.
Misregulated mTORC2 activity results in altered cell motility in various cell types, including different cancer cells where migration plays an important role in metastasis (Dada et al., 2008; Liu et al., 2006; Masri et al., 2007). Overall, the molecular mechanism by which mTORC2 regulates the actin cytoskeleton remains unclear.
In 2005, the first direct substrate of mTORC2, Akt, was identified. mTORC2 was found to be a long sought-after kinase phosphorylating Ser473 in the hydrophobic motif of Akt (Figure I- 3f) (Hresko and Mueckler, 2005; Sarbassov et al., 2005). Although earlier knockdown studies of rictor showed decreased phosphorylation also of Thr308 in the activation loop, further studies in knockout mice suggested that phosphorylation of Thr308, by phosphoinositide-dependent kinase 1 (PDK1), does not depend on prior Ser473 phosphorylation (Frias et al., 2006; Guertin et al., 2006; Jacinto et al., 2006; Shiota et al., 2006). The independent phosphorylation of Thr308 and Ser473 contrasts with the hierarchical phosphorylation of Thr229 and Thr389 in S6K; phosphorylation by mTORC1 (Thr389) is required for subsequent phosphorylation by PDK1 (Thr229) (reviewed in (Mora et al., 2004)). Furthermore, rather than being inactive, Akt without Ser473 phosphorylation appears to remain largely active as determined by the phosphorylation state of the Akt substrates glycogen synthase kinase 3 (GSK3), Tuberous Sclerosis Complex protein 2 (TSC2), Bad and the forkhead class O transcription factors 1/3a (FoxO1/3a). Only FoxO1/3a and possibly Bad show decreased phosphorylation upon loss of Ser473 phosphorylation (Guertin et al., 2006; Jacinto et al., 2006; Yang et al., 2006). Thus, mTORC2-mediated Akt phosphorylation does not seem to determine absolute activity, but rather appears to
determine substrate specificity. It is also possible that under conditions of low Akt activity, some Akt substrates can be phosphorylated by another kinase. For example, under conditions of insulin resistance when Akt is no longer active, GSK3 is phosphorylated by S6K (Zhang et al., 2006). Furthermore, it has been shown that FoxO can also be phosphorylated by SGK1 (Brunet et al., 2001), another mTORC2 substrate, providing a possible explanation for why cells with reduced mTORC2 activity show reduced FoxO phosphorylation but not reduced GSK3 phosphorylation.
Following the identification of Akt as an mTORC2 substrate, other AGC kinases were identified as additional substrates. In particular, many groups focused on the phosphorylation of PKC. Sarbassov et al. showed that PKCα phosphorylation (at the hydrophobic motif) and activity depend on mTORC2 (Sarbassov et al., 2004). However, this study suggested that the control of PKCα by mTORC2 is indirect. Ikenoue et al. (Ikenoue et al., 2008) and Facchinetti et al. (Facchinetti et al., 2008) later showed that mTORC2 is required for phosphorylation of all conventional PKCs and the novel PKCε at their hydrophobic motif and, in addition, at their turn motif, thereby controlling post-translational processing and stability of PKC (Figure I-3f). Loss of mTORC2 activity results in a reduction in total protein levels of PKC. Ikenoue et al. and Facchinetti et al., also showed that mTORC2 directly phosphorylates the turn motif in Akt. Interestingly, only phosphorylation of the hydrophobic motif, but not the turn motif, of PKC and Akt occurs in a growth-factor dependent manner. Overall, whether mTORC2 is the direct kinase of PKC remains an open question, as no study to date has been able to demonstrate direct in vitro phosphorylation of either site on any PKC isoform by mTORC2. Furthermore, it remains unclear how strongly mTORC2 activity influences PKC-mediated signaling events.
As discussed above, YPK2 and Gad8 were identified early on as TORC2 substrates, in budding and fission yeast, respectively. Both AGC kinases have close homology to the mammalian SGK kinase family. However, SGK1 was identified in mammals as an mTORC2 substrate only very recently (Garcia-Martinez and Alessi, 2008). mTORC2 phosphorylates SGK1 at its hydrophobic motif site (Figure I-3f) and thereby regulates SGK1’s activity toward its physiological substrate n-myc downstream regulated 1 (NDRG1). Given the very modest reduction in Akt activity upon loss of mTORC2, as discussed above, is Akt a major mTORC2 effector? Studies in both yeast and worms suggest that SGK is the main TORC2 effector in these organisms. Whether SGK1 is also the most important physiological substrate of TORC2 in mammals is not clear. SGK1-SGK3 double knockout (DKO) mice have a mild phenotype, including a defect in renal function that does not affect embryonic development (Grahammer et al., 2006). In contrast, Akt1-Akt2 DKO mice and Akt1-Akt3 DKO mice are
impaired in development, and the latter display a phenotype similar to that of rictor knockout mice (Guertin et al., 2006; Peng et al., 2003; Shiota et al., 2006; Yang et al., 2005).
Furthermore, loss of either rictor (Guertin et al., 2009) or Akt1 (Chen et al., 2006) suppresses the development of prostate neoplasia in Pten (phosphatase and tensin homolg) deficient mice. These findings are similar to those in D. melanogaster where tissue hyperplasia and increased Akt activity induced by Pten loss are reduced upon loss of rictor. Curiously, loss of rictor in the Pten prostate cancer mouse model reduces Akt phosphorylation at both Thr308 and Ser473. Overall, Akt still seems to be an important mTORC2 effector, at least upon enhanced signalling through the PI3K pathway.
While the processes downstream of TORC2 are coming into focus, knowledge on TORC2’s upstream regulators is almost completely lacking. In yeast, absolutely nothing is known about extracellular or intracellular signals controlling TORC2. In mammalian cells, mTORC2 phosphorylates Akt upon serum stimulation, in particular growth factors such as insulin and insulin-like growth factor 1 (IGF1), suggesting that mTORC2 is regulated by the PI3K pathway (Frias et al., 2006; Ikenoue et al., 2008; Jacinto et al., 2006; Yang et al., 2006).
This observation alone does not indicate that intrinsic mTORC2 kinase activity is stimulated by the PI3K pathway. Activation of PI3K leads to the production of phosphatidylinositol 3,4,5 trisphosphate (PIP3) and recruitment of Akt to the plasma membrane where it is phosphorylated by PDK1 and a possibly constitutively active, membrane-bound mTORC2.
Thus, mTORC2 could be constitutively active and its regulated phosphorylation of Akt is controlled at the level of Akt localization. However, arguing against this possibility and in favor of a model in which the PI3K pathway stimulates intrinsic mTORC2 kinase activity are the observations that mTORC2-dependent mTOR autophosphorylation at Ser2481 (Copp et al., 2009) and in vitro mTORC2 activity are stimulated by growth factors (Frias et al., 2006;
Yang et al., 2006). Furthermore, mTORC2 appears to phosphorylate SGK1 in response to growth factors even though SGK1 lacks a PH domain and is activated independently of membrane recruitment. Taken together, these latter findings suggest that growth factors, via the PI3K pathway, stimulate intrinsic mTORC2 kinase activity.
How might growth factors activate mTORC2 kinase activity? Despite several indications that growth factors stimulate mTORC2 activity, it remains a mystery how the growth factor signal is relayed within the cell to activate mTORC2. A recent report suggests that growth factors could signal to mTORC2 via the TSC1-TSC2 complex (a complex of the two tuberous sclerosis complex proteins 1 and 2). Huang et al. (Huang et al., 2008) have proposed that the TSC1-TSC2 complex, a GTPase activating protein (GAP) that lies upstream of and negatively regulates mTORC1, also regulates mTORC2 function by binding directly to mTORC2. In contrast to the negative regulation of mTORC1 by TSC1-TSC2,
TSC1-TSC2 is proposed to positively regulate mTORC2 activity in a GAP-independent manner. The GTPase Rheb, which lies directly downstream of TSC1-TSC2 and activates mTORC1 (Manning and Cantley, 2003), does not appear to lie upstream of mTORC2. The observation that TSC1-TSC2 GAP-activity is not required for mTORC2 activation suggests that the mechanism is not via mTORC1 and the negative feedback loop, which is a hallmark of activated mTORC1 signaling. The mTORC1 substrate S6K directly phosphorylates insulin receptor substrate (IRS) thereby promoting degradation of IRS (Tremblay et al., 2007). As a consequence, further Akt signaling by insulin is attenuated and cells are in an insulin- resistant state. Indeed, Huang et al. argue that TSC-mediated activation of mTORC2 is not via the negative feedback loop. The way in which TSC1-TSC2 binding to mTORC2 regulates mTORC2 activity as well as potential GAP-independent activities for TSC1-TSC2 remain poorly understood.
Is mTORC2 found at the plasma membrane and is mTORC2 localization regulated?
In the Pten prostate cancer mouse model, where PI3K signaling is increased, rictor and Ser473-phosphorylated Akt are enriched at the plasma membrane. Furthermore, Partovian et al. (Partovian et al., 2008) demonstrated that a Syndecan-4 deficiency reduces mTORC2 localization to detergent-insoluble membrane fractions (rafts) in endothelial cells.
Interestingly, Syndecan-4, which is a single-pass transmembrane proteoglycan, recruits PKCα to the plasma membrane and thereby regulated PKCα activity; this, in turn, is required for proper mTORC2 localization to the rafts and subsequent Akt activation. However, the mechanism by which PKCα regulates mTORC2 recruitment is not known, and is further complicated by the fact that PKCα is known to be a downstream target of mTORC2.
Although we have some insights into how mTORC2 might be regulated, it will be a major breakthrough to identify upstream regulators of the TORC2 signaling branch, in both yeast and mammals.
Concluding remarks and future perspectives
We have summarized and highlighted the major recent findings on TORC2 in various organisms. The sum of all studies clearly shows conserved functions of TORC2 across organisms (Figure I-1). In plants and algae, many components of the TOR signaling pathway have been elucidated and Arabidopsis Thaliana TOR (AtTOR) is important in the control of plant growth (Anderson et al., 2005; Deprost et al., 2007; Diaz-Troya et al., 2008;
Mahfouz et al., 2006). However, no rictor or Sin1 homologue has been identified in A.
thaliana or the green alga Chlamydomonas reinhardtii questioning the existence of TORC2 in photosynthetic organisms. Interestingly, TORC2 was recently identified in the protozoan parasite Trypanosoma brucei, which causes sleeping sickness in humans (Barquilla et al.,
2008). In contrast to other eukaryotes, rapamycin treatment of T. brucei inhibits cell growth by exclusively preventing TORC2 assembly, without affecting TORC1.
TORC2, together with its sibling complex TORC1, is the main kinase that phosphorylates and thereby regulates the activity of several AGC kinases. TORC2 also regulates actin cytoskeletal organization in most systems studied. However, it remains unclear how TORC2 specifically regulates this process in the context of diverse physiological processes that involve motility, such as embryogenesis, inflammation, metastasis, or wound healing.
Several studies have shown that a full body knockout of any mTORC2 component is embryonic lethal. As a next step, it would be highly interesting to determine how mTORC2 in individual organs influences whole body growth and metabolism. Conditional knockout studies of the mTORC2-specific component rictor in skeletal muscle display minimal phenotypes (Bentzinger et al., 2008; Kumar et al., 2008), suggesting that the role of mTORC2 in muscle is less important. Loss of rictor in adipose tissue, however, results in a more dramatic phenotype. Adipose mTORC2 negatively controls whole-body growth and also influences glucose metabolism by influencing IGF1 and insulin levels, respectively (work of this thesis: Cybulski et al., 2009).
Does mTORC2 have a role in diseases such as cancer or metabolic disorders? Guertin et al. (Guertin et al., 2009) made the interesting observation that mTORC2 is important for the development of prostate cancer induced by Pten loss, but is not important for non-cancerous prostate epithelial cells. Moreover, as conditional knockout studies show a role for mTORC2 in regulating glucose metabolism, mTORC2 could play a role in the development of type 2 diabetes ( work of this thesis: Cybulski et al., 2009; Kumar et al., 2008). The involvement of mTORC2 in diseases is only starting to be considered. Future studies might reveal the need for drugs that specifically inhibit mTORC2, such as rapamycin for mTORC1.
mTOR associated diseases
Dysregulated mTOR signaling is often linked to tumor formation (Guertin and Sabatini, 2007). mTOR itself is not mutated in tumors or cancers, but many upstream regulators and downstream effectors that are functionally linked to the mTOR signaling pathway are well- known proto-oncogenes or tumor suppressors. The most evident link between aberrant mTOR signaling and tumor formation is found in patients with tuberous sclerosis complex, a disease characterized by hamartomas (begin tumors) caused by inactivating mutations of the
tumor suppressors TSC1 and TSC2 (Inoki et al., 2005). Hamartomas are also found in patients with Peutz-Jeghers syndrome (PJS). In PJS mTOR signaling is increased due to a loss of function mutation in the negative upstream regulator LKB1. Further proto-oncogenes that are commonly activated in cancers and that result in increased mTOR signaling are PI3K, Akt/PKB, Rheb and S6K1, while inactivating mutations in the tumor suppressors PTEN and 4E-BP1 have the same effect (Wullschleger et al., 2006). So far, the direct link between cancer and mTOR is limited to the mTORC1 signaling branch and rapamycin and its derivatives have already been tested in clinical studies for cancer treatment. However, little is known about the role of the mTORC2 signaling branch in cancer. First indications that mTORC2 signaling can be misregulated in cancers come from studies that show that mTORC2 is required for the development of prostate cancer caused by PTEN deletion (Guertin et al., 2009).
Several studies in mice have shown that mTORC1 signaling in various metabolic tissues plays an important role in regulating whole body energy metabolism (Polak and Hall, 2009) and dysregulation of mTOR signaling is associated with metabolic disorders such as obesity and diabetes. As described above, constitutive activated mTORC1 signaling inhibits IRS via the negative feedback loop from mTOR-S6K resulting in a strong inhibition of the insulin- mediated PI3K pathway. As a consequence, cells become desensitized to insulin, causing insulin resistance.
Far less is known about mTORC2 and its role in regulating whole body energy metabolism. First knockout studies revealed that, at least in muscle, loss of mTORC2 results in little-to-no phenotype. Muscle-specific rictor knockout mice are slightly glucose tolerant but don’t show any further metabolic abnormalities (Bentzinger et al., 2008; Kumar et al., 2008).
Adipose tissue
It was believed for a long time that the adipose tissue solely functions as a fat storage compartment within the body. However, during the last few years intensive research has replaced this view by the notion that adipose tissue also has a central role in lipid and glucose metabolism and functions as an endocrine organ. Adipose tissue is an important player in the regulation of energy homeostasis (Figure I-4).
Adipose tissue is considered to be a vital organ, is unique to vertebrates and is found in most mammals, birds, reptiles and amphibians. It is mainly composed of the fat-storing adipocytes, which can determine up to 85% of the white adipose tissue mass, but adipose tissue also comprises other cell types, such as macrophages, fibroblasts, preadipoyctes, nerve and endothelial cells. Adipocytes have a great capacity to store fat and can expand to a size larger than most other cell types. The fat is stored within a single large lipid droplet that can represent approximately 90% of the cell volume of an adipocyte (Haugen and Drevon, 2007). In addition to white
adipose tissue, also brown adipose tissue exists. Brown adipocytes store fat in multiple lipid droplets and they contain a large number of mitochondria, which give them their brown appearance. In contrast to the fat storage function of white adipose tissue, brown adipose tissue has a primary function in heat generation and adaptive thermogenesis by means of uncoupling of the proton gradient from ATP production. Due to this function brown adipose tissue has an important role in the thermoregulation of small, hibernating rodents. In humans, brown adipose tissue is found in newborns but is largely replaced by white adipose tissue in adults (Cannon and Nedergaard, 2004).
In the following part the role of white adipose tissue in fat storage, fat release, its function as an endocrine organ, as well as associated diseases will be described in more detail.
Figure I-4: Interplay of metabolic tissues.
Adapted from Shi and Burn, 2004.
Fat storage and release
The primary role of adipose tissue, in particular of the adipocyte cell, is to store energy in the form of triacylglycerol (TAG) during times when input exceeds expenditure and to break down this stored lipid into free fatty acids (FFA) when energy is required (Haugen and Drevon, 2007). TAGs are lipid molecules composed of three fatty acids attached to a glycerol backbone. Fatty acids can be saturated or unsaturated. The main source of FFAs comes from the diet, but upon shortage, FFAs can also be synthesized by the body from glucose in a process called de novo lipogenesis. In the diet, fat is mainly stored as TAGs, and many fats that originate from animals are mainly composed of saturated fatty acids, whereas vegetable-derived fats often contain a high percentage of unsaturated fatty acids.
Upon food intake, dietary fats are first digested in the stomach and upper small intestine through the action of stomach acids, bile salts, and digestive enzymes, such as lipases. In the intestinal cells, the absorbed fat is re-assembled into TAGs and packaged together with cholesterol, fat-soluble vitamins and carrier lipoproteins into small particles called chylomicrons that are then secreted into the blood stream. Except for the liver, most organs are able to extract the lipids from the chylomicrons with the help of lipoprotein lipases, which are secreted by the individual organs and are found in the capillaries of mainly heart, skeletal muscle and adipose tissue, but also other non-liver organs. Lipoprotein lipases break the TAGs down into free fatty acids. While the heart and skeletal muscle metabolize the fatty acids for energy production, the adipose tissue stores the fatty acids by forming new TAGs.
Once the chylomicrons are depleted of TAGs and remain loaded with cholesterol, they are known as chylomicron remnants that are taken up by the liver (Shi and Burn, 2004).
Between meals, the liver can also synthesize TAGs from glucose, which are then distributed to the body in the form of very low density lipoproteins (VLDL). VLDLs provide fatty acids mainly to the heart and skeletal muscle for oxidation and as more and more TAGs are removed, the composition of the VLDL changes and it becomes an intermediate-density lipoprotein (IDL). Further loss of TAGs results in smaller and denser particles called low density lipoproteins (LDL), that are then enriched in cholesterol. LDLs provide cholesterol to peripheral tissues. LDL is often referred to as the ‘bad’ cholesterol fraction, since an excess of LDL in the blood stream is a major cause of atherosclerosis. The ‘good’ cholesterol fraction is known as high density lipoprotein (HDL). HDL is assembled in the blood and transports excess cholesterol away from or out of tissues back to the liver for eventual disposal.
As mentioned before, the adipose tissue stores dietary fat as TAG. FFAs that are released from lipoproteins (mainly chylomicrons) by lipoprotein lipase enter the adipocytes
through both passive diffusion and active transport. FFAs are toxic to cells, so once inside the cell they are transported via fatty acid-binding proteins and converted to acyl coenzyme A (acyl-CoA). This activated form of FFA can then be either oxidized by mitochondria for energy generation (more common in other cells), or be transported to the endoplasmic reticulum for esterification with glycerol 3-phosphate that is generated by glucose metabolism. The formation of TAGs requires several enzymatic steps and once synthesized the TAGs are stored away into lipid droplets.
During fasting or exercise, fatty acids and glycerol are released from the adipose tissue into the blood stream in order to supply peripheral tissues with sufficient energy. The process of lipid mobilization by releasing FFAs is called lipolysis. In the fed state, lipolysis is inhibited by insulin, while in the starved state epinephrine and other lipolytic hormones promote the breakdown of stored TAGs. At least three adipocyte lipases are required to release the three fatty acids bound to one glycerol backbone. Adipose TAG lipase (ATGL) removes the first fatty acid, followed by hormone-sensitive lipase (HSL) that degrades the diacylglycerols into monoacylglycerols. The last step requires monoglyceride lipase.
Activation of lypolysis through lipolytic hormones causes an increase in intracellular cyclic AMP (cAMP) levels and activation of cAMP-dependent protein kinase A (PKA). Activated PKA phosphorylates and activates HSL, which regulates the rate-limiting step of lipolysis.
PKA also phosphorylates perilipin, a lipid-droplet-coating protein, that when phosphorylated, facilitates access of HSL to the lipid droplet, and hence lipolysis. Once released into the blood, FFAs are transported in an albumin-bound form to different tissues that can oxidize them, mainly liver, heart, kidney and muscle (Shi and Burn, 2004).
Endocrine functions
Adipose tissue has long been believed to be solely a fat storage depot, but it is now clear that adipose tissue is a complex and highly active metabolic and endocrine organ. It has the capacity to actively communicate by sending and receiving different types of signals.
Adipose tissue expresses many receptors that allow it to respond to endocrine and autocrine signals, such as insulin, glucagon, catecholamines, insulin like growth factor 1 (IGF1) and many others. On the other hand, it also secretes a variety of factors, including metabolites and proteins that derive from both adipocyte and nonadipocyte fractions. These metabolically active adipose-derived factors are commonly known as “adipokines” and they affect whole body energy metabolism and the function of many organs and tissues including
muscle, liver, vasculature, and brain. Adipokines act in an autocrine, paracrine and/or endocrine manner and are involved in glucose and lipid metabolism, inflammation, coagulation, blood pressure and feeding behavior (Kershaw and Flier, 2004; Wang et al., 2008).
The two best-characterized and most studied adipokines are leptin and adiponectin.
Leptin is a small protein secreted from adipocytes and its serum levels correlate with adipose tissue mass as well as the nutritional status (Friedman, 2009). However, leptin levels are not directly influenced by short-term nutritional status (food intake) but rather reflect the long- term nutritional status. Leptin acts through the sympathetic nervous system to negatively regulate appetite and food intake. Its binding to the leptin receptor in the hypothalamus results in the reduction of neuropeptide-Y (NPY) and agouti regulated protein (AgRP) activity, two orexigenic neuropeptides, while increasing the activity of the anorexigenic neuropeptides pro-opiomelanocortin (POMC) and cocaine- and amphetamine related protein (CART).
Leptin was discovered in 1994 as the product of the obese (ob) gene. Mice with a mutation in the ob gene are extremely obese due to the lack of functional leptin, resulting in a reduced feedback signal to stop eating. Besides its central role in regulating food intake and energy expenditure, leptin also influences lipid metabolism by increasing hepatic lipid oxidation and lipolysis in skeletal muscle and adipocytes (Hajer et al., 2008).
Adiponectin is another protein hormone that is produced exclusively by adipocytes and is regarded as an insulin sensitizer that plays an important role in the development of insulin resistance. In contrast to most other adipokines, serum adiponectin levels negatively correlate with adipose tissue mass (Garaulet et al., 2007). Adiponectin improves insulin sensitivity by reducing hepatic glucose production and enhancing insulin action in the liver and skeletal muscle.
Many additional adipokines and their associated role in regulating whole body energy metabolism have been identified (Kershaw and Flier, 2004; Wang et al., 2008). The following table gives an overview of the functional groups of adipokines and the major factors involved:
Functional category Factors
Lipid metabolism Cholesteryl ester transfer protein (CETP), Lipoportein lipase (LPL), Retinol binding protein 4, Apolipoprotein E, Steroid hormones Glucose metabolism and
insulin resistance
Adiponectin, Resistin, Visfatin, Omentin, Vaspin, Leptin, TNFα
Food intake Leptin
Inflammation Tumor necrosis factor α (TNFα), Interleukin-6 (IL-6), Adiponectin, Resistin, C-reactive protein (CRP), Adipsin
Vasculature Angiotensin, Vascular endothelial growth factor (VEGF), Adrenomedullin, Plasminogen activator inhibitor-1 (PAI-1)
Adipose tissue associated diseases
Body fat has the very important function of storing fatty acids for times of low energy availability. However, the right amount of body fat is very important as too much or too little fat are a serious health risk.
Obesity, an excess of body fat results from an imbalance of caloric intake and energy expenditure, but also genetic factors can influence the progression of obesity. Obesity is strongly associated with many diseases, in particular diabetes and cardiovascular diseases (Hajer et al., 2008). As mentioned before, adipose tissue has an important function as an endocrine organ to control whole body energy homeostasis and obesity is often associated with marked changes in the secretory function of the adipose tissue that promotes the development of diseases.
Obesity-linked type 2 diabetes is due to a combination of insulin resistance and dysfunction of the insulin-secreting pancreatic β-cells (Guilherme et al., 2008; Lingohr et al., 2002). Insulin resistance results from an impaired insulin responsiveness of skeletal muscle resulting in diminished glucose uptake. In a first step, normal glucose levels can be maintained by increased insulin production/secretion by pancreatic β-cells. However, when the pancreatic β-cells fail to secrete enough insulin to compensate for insulin resistance, blood glucose levels rise and diabetes will ensue. Several adipose-derived factors, such as adiponectin, TNFα, leptin, IL6 and FFAs, are known to affect insulin sensitivity of peripheral tissues, and they furthermore have effects in the pancreas leading to β-cell failure (Hajer et al., 2008).
Obesity is also a major risk factor for hypertension and cardiovascular diseases, in particular coronary heart disease (Bays, 2009). High blood glucose levels (resulting from type 2 diabetes), elevated blood pressure, elevated TAGs, low plasma HDL and high plasma LDL are associated with obesity and are all known to increase the risk for coronary heart disease.
